MOSHPIT - questions about contigs table and visualization outputs from contigs QC step

[blueridge691]

Hello,

I am loving the functionality of MOSHPIT, parsl, and the artifact cache

Re: MOSHPIT, I do have a few questions about outputs from Contigs QC step (see script excerpt below):

# Contigs QC
mosh assembly evaluate-quast \
  --i-contigs "$CACHE_DIR:contigs" \
  --p-threads 16 \
  --p-memory-efficient \
  --o-visualization ./results/contigs.qzv \
  --parallel-config "$PARSL_CONFIG" \
  --use-cache "$CACHE_DIR" \
  --o-results-table ./results/contigs_results_table.qza \
  --o-reference-genomes "$CACHE_DIR:reference_genomes" \
  --verbose

wait
echo "Finished 2 of 11 steps"

1. Is there a way to regenerate the contigs_results_table.qza with my existing contigs? Can I merge 2 or more contig tables?
2. My contigs.qzv output file was multiple GBs and understandably did not load in QIIME 2 View. Is there another way to extract the information contained in this visualization? For context, I was running 80+ metagenomic samples through this pipeline in parallel.

Thanks so much for all you do!

Sincerely,

Lexie

[misialq (Michal Ziemski)]

Hey @blueridge691,

welcome to the forum! We are very happy to hear you like all the functionality! Now, about your questions:

1. Do you mean running the evaluate action on contigs that were assembled outside of MOSHPIT? If yes, you could import them into a new SampleData[Contigs] artifact (see here for details). Once you have them in your artifact you can run the evaluate action as usual.

   About merging the results: it is not possible to merge multiple QUASTResults into a new QUASTResults artifact. However, you could pass those tables to the merge action from the metadata plugin like this:
   qiime metadata merge --m-metadata1-file contigs_results_table1.qza --m-metadata2-file contigs_results_table2.qza --o-merged-metadata merged_contigs_results.qza
   This will create a new artifact which is of the ImmutableMetadata type so you can still use it in all the actions which expect you to input metadata (e.g., if you wanted to filter-contigs).

2. Hmmm, that's a bummer - perhaps we need to investigate a bit why those visualizations get so large In the meantime, you could export the data out of it by using the qiime tools export action. This will unpack all of the files into a new directory which you can then browse as usual. You should find in there a bunch of TSV files with different result tables and pdfs with the Nx and GC plots. Perhaps you could also try to run the evaluate action with the --p-no-icarus flag - this will tell QUAST to not draw the Icarus contig browser plots and help reduce the size a bit (I cannot tell you, though, by how much exactly).

I hope that helps! Let us know if anything is still unclear

Best,
Michal

[blueridge691]

Hi @misialq,

Thanks for the quick response! I have a clear idea of next steps

Much appreciated!